InvestIndependently confirmed2 publishers2 min readPublished
Intel's Tan calls DeepSeek, Kimi and GLM 'good enough' while Washington holds 9.9% of his company
Intel CEO Lip-Bu Tan called DeepSeek, Kimi and GLM "good enough" and cheap to run at a Singapore panel on Monday. Washington owns 9.9% of Intel and named the same labs in a September advisory, so the cost case comes from a company it part-owns.
The Investor · Invest desk

Key figures: the US government agreed to buy 433.3 million Intel shares at $20.47 each, an $8.9 billion investment, per Intel's filing. Counting earlier CHIPS Act grants, total US government backing of Intel reaches $11.1 billion.
| Figure | Value | Note | Claim |
|---|---|---|---|
| US government's agreed investment in Intel shares | 8.9 $ billion | 10 | |
| Intel shares to be bought | 433.3 million shares | 10 | |
| Price per share | 20.47 $ | 10 | |
| Total US government backing of Intel | 11.1 $ billion | counting earlier CHIPS Act grants | 11 |
What happened
- A September advisory from the FBI, NSA and CISA named DeepSeek, Moonshot AI and Z.ai among six Chinese firms running industrial-scale distillation of US models since late 2024.
- Under Intel's deal with the Trump administration, the government bought 433.3 million shares at $20.47 each, an $8.9 billion investment, per Intel's filing.
- Tan spoke on a Singapore panel with Broadcom's Hock Tan and named the cost of capital and uncertain returns as the AI industry's key challenges.
- Nvidia's Jensen Huang said in July that US companies should "absolutely" be allowed to use Chinese open-source models, which he called "excellent".
Why it matters
- contradiction Washington's agencies accuse three labs while a company it part-owns praises them, and Trump's suggestion that the government "might" take stakes in OpenAI and Anthropic would put it on both sides.
- cost Where a Chinese open-weight model is good enough, the premium on a US closed model becomes a cost the buyer must justify, and US closed-model labs are the ones asked to defend it.
- decision Policymakers must choose between acting on the reported ban and sanction threats and leaving buyers to choose on price, a step the July letter's signatories, including Nvidia and Microsoft, asked them not to rush.
The government's position in Intel works out to 433.3 million shares at $20.47, or $8.87 billion [15] (the filing rounds it to $8.9 billion [10]), and the $11.1 billion total [11] implies about $2.2 billion of earlier CHIPS Act grants [16]. That is a large holding in a company whose chief executive has just endorsed three labs the same government's agencies accused of distilling US models [8].
We think the remark is a vendor's demand argument. Intel sells chips, and Nvidia's Jensen Huang made the commercial version in July, arguing, according to Cryptopolitan, that cheaper models expand demand for chips [14]. Tan's own words fit that reading: "I'm a strong believer of open source, that's the best way to drive efficiency," he said [3]. The counter-thesis is that the stake changes how the remark is heard, and Cryptopolitan calls the endorsement politically loaded for that reason [19].
Ctech's account of the same panel puts the remark inside a conversation about financing. "If there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the long run, people want to see the ROI," Lip-Bu Tan said [5]. Broadcom's Hock Tan argued on the same panel that financing could become a bigger constraint than chips, memory or electricity [6], and said demand for compute could outrun supply through 2029 and potentially into 2030 [7]. A buyer who has to show a return on expensive capital has a reason to ask whether a model that is good enough for some uses [2] also costs a fraction of the closed alternative [17]. Ctech says the remark challenges the assumption that the most expensive systems will dominate every market [18].
The evidence stops short of the policy claim. Tan said "good enough" for some uses [2], and Neither article reports a price per token or any response from the administration.
The policy side is where the record is thinnest. In July, Nvidia, Microsoft, Meta, Hugging Face and Mistral signed a letter against "premature restrictions" on open-weight models, and it coincided with reports of ban and sanction threats from the Trump administration and Treasury Secretary Scott Bessent [13]. If the administration acts on those threats, Intel's chief executive will have publicly endorsed models the administration then restricted. If buyers keep Chinese models to the narrow uses Tan described and US closed models keep the rest, the pressure stays a pricing matter for US labs and never reaches policy.
What to watch
- Any Treasury or Commerce move on Chinese open-weight models, given the ban and sanction threats reported around July.
- A public response from the administration or Intel to Tan's remarks, or a formal government stake in OpenAI or Anthropic.
- Price-per-token or workload-share disclosures from US buyers that would show whether the 'some uses' Tan described are widening.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Insufficient
- Hype gap+20
- Incentives
- Insufficient
- Confidence60
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Intel CEO Lip-Bu Tan on Monday backed Chinese open-weight models DeepSeek, Kimi and GLM, calling them 'good enough' and cheap to run.
- [2]
Tan said the token costs of Chinese frontier AI models have been lower and their adoption 'very fast', and that models including DeepSeek, Kimi and GLM are 'good enough' for some uses.
- [3]
"I'm a strong believer of open source, that's the best way to drive efficiency," Tan said.
ReportedSupportedSource: Lip-Bu Tan, quoted by Ctech (Calcalist)2 sources— create a free account to open themView cited source - [4]
Tan spoke on a panel with Broadcom CEO Hock Tan at the DBS Asia Future Forward Conference in Singapore, naming the rising cost of capital and uncertainty over returns on investment as key challenges for the AI industry.
- [5]
"If there's any challenge in the front, it's capital and the cost of capital, and then the other part is in the long run, people want to see the ROI," Lip-Bu Tan said.
- [6]
Broadcom CEO Hock Tan argued that capital could become a bigger obstacle to AI expansion than shortages of chips, memory or electricity.
- [7]
Hock Tan said demand for computing infrastructure exceeds what the industry can supply, a gap that could persist through 2029 and potentially into 2030.
- [8]
A September joint advisory from the FBI, NSA and CISA accused DeepSeek, Moonshot AI (Kimi), Z.ai (GLM) and three other Chinese firms of running industrial-scale distillation campaigns against US models since late 2024.
- [9]
Distillation, training a smaller model on a larger one's outputs, is a standard research method, but Washington has pushed back at the scale and intent of the Chinese campaigns.
- [10]
Under a deal with the Trump administration, the US government agreed to buy 433.3 million Intel shares at $20.47 each, an $8.9 billion investment representing a 9.9% stake, according to Intel's filing.
- [11]
Counting earlier CHIPS Act grants, total US government backing of Intel reaches $11.1 billion.
- [12]
In July Nvidia CEO Jensen Huang said US companies should 'absolutely' be allowed to use Chinese open-source models, which he called 'excellent', and dismissed fears that Chinese AI could run US firms 'off the road'.
- [13]
In July Nvidia joined Microsoft, Meta, Hugging Face and Mistral as signatories of an open letter asking policymakers not to rush to 'premature restrictions' on open-weight models; the letter never named China but coincided with reports of ban and sanction threats from the Trump administration and Treasury Secretary Scott Bessent.
- [14]
Huang argued, in a July interview with Axios, that cheaper models expand demand for chips.
- [15]
433.3 million shares at $20.47 each comes to about $8.87 billion.
- [16]
The gap between $11.1 billion of total backing and the $8.9 billion share purchase implies about $2.2 billion of earlier CHIPS Act grants.
- [17]
Developers can download and adapt open-weight models, including those from the Chinese labs Tan named, at a fraction of the price of running similar-performing closed models from American labs.
- [18]
Ctech wrote that Tan's remarks challenge the assumption that the most expensive systems will necessarily dominate every market.
ReportedInsufficientSource: Ctech (Calcalist)2 sources— create a free account to open themView cited source - [19]
Cryptopolitan describes Tan's endorsement as politically loaded because the US government holds about 10% of Intel.
ReportedInsufficientSource: Cryptopolitan2 sources— create a free account to open themView cited source - [20]
Trump has suggested the government 'might' take Intel-style stakes in OpenAI and Anthropic, the two labs loudest about Chinese distillation, which would leave Washington owning both the chipmaker praising Chinese models and the labs accusing them of theft.
Sources
2 independent publishers whose own reporting we read for this story.
- calcalistech.comIntel’s Lip-Bu Tan says AI companies will have to prove the returns justify the spend | Ctech
1 article · October 12, 2026
- cryptopolitan.comIntel’s Tan joins Nvidia in defending Chinese open-weight AI
1 article · October 12, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- Open-Weight AI ModelsFollow
- US government equity in IntelFollow
- US-China AI PolicyFollow
- AI financing and returnsFollow